Professional Documents
Culture Documents
Ultra-Small-Angle Neutron Scattering With Azimuthal Asymmetry
Ultra-Small-Angle Neutron Scattering With Azimuthal Asymmetry
a
Department of Geosciences, Pennsylvania State University, University Park, PA 16802, USA, and bNIST Center for
Neutron Research, National Institute of Standards and Technology, Gaithersburg, MD 20899, USA. *Correspondence
e-mail: david.mildner@nist.gov
1. Introduction
The typical range of scattering vectors measured on a pinhole
small-angle neutron scattering (SANS) instrument is 103 <
1, suitable for measuring correlations and dimenQ < 0.5 A
. The modulus of the scattering
sions in the range 101000 A
vector is defined by Q = (4/) sin (/2), or Q 2/ at small
angles, where is the wavelength and is the scattering angle,
and the measured dimensions are in the range of Q1. The
circular isointensity contours on an area detector for isotropic
systems that scatter with azimuthal symmetry may be
azimuthally averaged to obtain a one-dimensional plot of the
scattered intensity I(Q) as a function of Q. Moreover, a twodimensional detector enables observation of the intensity
from azimuthally asymmetric scattering samples. Efforts are
underway to extend the lower end of the Q range by a further
order of magnitude using pinhole geometry, in order to
examine dimensions as large as 1 mm (Brulet et al., 2008;
Dewhurst, 2014; Barker, 2015; Abbas et al., 2015).
For systems having dimensions greater than 1 mm, an
alternative method is required to extend the measurable range
to lower values of the scattering vector. Ultra-small-angle
neutron scattering (USANS) using a double-perfect-crystal
diffractometer can measure in the scattering vector range of
1 and examine dimensions in the range
105 < Q < 103 A
0.110 mm (Agamalian et al., 1997; Barker et al., 2005). This
instrumental arrangement is different in concept from pinhole
SANS in that it measures with high resolution in one direction
only, while having a much broader angular acceptance in the
934
http://dx.doi.org/10.1107/S1600576716005586
research papers
orthogonal direction in order to increase the count rate. This
has been made possible by improvements to the BonseHart
arrangement (Bonse & Hart, 1965) of silicon channel-cut
crystals by the elimination of the wings on the rocking curve
(Agamalian et al., 1997).
The smearing of the scattered intensity for USANS is
similar to slit-smeared small-angle X-ray scattering. If a model
that describes the scattering system exists, the smeared scattering function should be compared with the USANS data.
However, if such a model is unavailable, it is necessary to
desmear the USANS data in order to place the scattered
intensity I(Q) on an absolute basis with the SANS data from a
conventional pinhole instrument. The desmeared USANS
intensity from systems that scatter with azimuthal symmetry
may conveniently be concatenated with the azimuthally
averaged data obtained from the accompanying SANS
measurement. This procedure is not available for anisotropic
systems and consequently there are few USANS data
analyzed on azimuthally asymmetric systems. We examine the
analysis of systems for which the scattering pattern consists of
elliptical isointensity contours.
Asymmetric data often have contours with an elliptical
dependence, which indicate that the scattering is from entities
that have azimuthal symmetry about some unique axis within
the system (Summerfield & Mildner, 1983). Ciccariello et al.
(2000) have generalized the Porod law as a function of the
direction of Q for anisotropic samples. The unique axis may be
a stretched direction as in extruded materials (such as the
uniaxial drawing of polyethylene; Matsuba et al., 2013), or it
may be caused by uniaxial compaction as in geological samples
(Gu et al., 2015). Sector averages over a range of azimuthal
angles are taken so that the scattered intensity I(Q) is
obtained with reasonable statistics along the two orthogonal
directions.
The desmearing of USANS data for conversion to pinhole
geometry uses an iterative method (Lake, 1967) that is good
for symmetric scattering data but not for asymmetric scattering because the two orthogonal dimensions are not
commensurate. We investigate the information that might be
gleaned from USANS data collected from such anisotropic
systems by taking measurements from rock samples in
different orientations.
2. Geological samples
USANS measures the scattering of neutrons from scattering
length density (SLD) fluctuations in real space in the micrometre range. It is particularly useful for systems in materials
science and polymer science that are optically opaque and
therefore cannot be examined by light scattering. The technique is also useful for the study of porous media such as
cements. Here, we consider a geological example to probe the
difficulties of treating asymmetric scattering systems. [For a
review, see Radlinski (2006)]. We assume that we have a twophase system with scattering from pores within the rock,
arising from the difference in SLD of the pores relative to
some average SLD of the rock matrix.
J. Appl. Cryst. (2016). 49, 934943
935
research papers
produce I(Q). Sector averaging over azimuthal angles
30 is necessary for the perpendicular-cut measurement
(averaging using 15 with fewer pixels results in noisier
data).
Similar USANS measurements were performed on the
double-crystal instrument BT5 (Barker et al., 2005) at the
with / = 5.9%. The
NCNR using a wavelength = 2.38 A
double-crystal arrangement has an instrumental (full width at
1 in the horizontal
half-maximum) resolution of 2 105 A
5
1, and
direction, with a Q range of 3 10 < Q < 3 103 A
with a relaxed resolution in the vertical y direction
1). Note that the perpendicular-cut sample was
( 0.117 A
first mounted with the bedding (direction Y) horizontal, in the
direction of the high-resolution USANS scan, and then
remounted so that the normal to the bedding (direction Z) was
in the Q scan direction. Both sets of measurements were
reduced to an absolute scale using the NIST data-reduction
procedure (Kline, 2006).
3. SANS results
Fig. 1 shows the two-dimensional SANS scattering patterns for
the samples cut parallel and perpendicular to the bedding
plane. We label the three principal directions, so that the
bedding plane is always the XY plane with the neutron beam
in the Z direction, and the perpendicular-cut sample defines
the YZ plane with the neutron beam in the X direction. The
two-dimensional SANS patterns are plots of I(Qx, Qy, 0) with
the neutron beam along Z, and I(0, Qy, Qz) with the neutron
beam along X.
IX Q IQ cos ; Q sin ; 0 0<<2 :
Figure 1
A schematic diagram indicating the planes of the two cuts, one parallel
and the other perpendicular to the plane of bedding of the sample rock.
The unique axis is the normal to the bedding plane. For each orientation
of the sample, the scattered neutron beam results in a pattern on the twodimensional detector. The scattering has circular (azimuthally symmetric)
contours when the beam is incident along the Z direction on the sample
cut parallel to the bedding plane XY, and elliptical (azimuthally
asymmetric) contours when the beam is incident along the X direction
on the sample cut perpendicular to the bedding plane YZ.
936
Gu and Mildner
Figure 2
The azimuthally averaged SANS results, with intensity plotted as a
1. The X curve is from the
function of Q over the range 0.001 < Q < 0.4 A
azimuthal average for the parallel-cut sample, the Y and Z curves are for
sector averages in directions parallel and normal to the bedding for the
perpendicular-cut sample, respectively, and the R curve is for the rotation
of the perpendicular-cut sample. Note that the Y curve is coincident with
the X curve since they are equivalent.
J. Appl. Cryst. (2016). 49, 934943
research papers
Table 1
1 with n =
The straight-line fits of the SANS data to a simple power law AQn over the scattering vector magnitude range 1 103 < Q < 2 102 A
m
3.33, together with the straight line fits of the smeared USANS data to a simple power law AQ over the scattering vector magnitude range 2 104 <
1 with m = 2.16.
Q < 2 103 A
SANS measurement
USANS measurement
2.21 (2)
2.15 (3)
14.26 (9)
7.11 (5)
6.00 (5)
9.86 (7)
20.25 (16)
15.76 (8)
IY Q I 0; Q sin ; Q cos =12<=2<=12
IZ Q I 0; Q sin ; Q cos =12<<=12
Figure 3
1 as a function of azimuth when (a)
The scattered intensity at Q = 0.02 A
the sample is cut parallel to the bedding with scattering in the XY plane,
and (b) the sample is cut perpendicular to the bedding with scattering in
the YZ plane. This shows that the Y measurement ( = /2) has a
similar intensity to the X measurement. The YZ data are fitted to
equation (4) so that I() [cos2 + (a/b)2sin2]n/2, where a/b = 1.76 and
n = 3.33.
J. Appl. Cryst. (2016). 49, 934943
Figure 4
A schematic diagram showing intensity I versus Q for the Y and Z
measurements, indicating that the two curves can be made coincident
either (i) by multiplying the Q coordinate for the Z data by b/a or (ii) by
multiplying the Z intensity values by (b/a)n.
Gu and Mildner
937
research papers
IR Q I 0; Q sin ; Q cos 0<<2
D
n=2 E n
a2 cos2 b2 sin2
Q :
hci =2 a E k;
sample. The figure also shows that the expected value for the
rotation measurement calculated from the Y and Z data
agrees well with the experimental results R.
4. USANS results
The double-perfect-crystal USANS instrument measures the
scattering for a given value of Q by summing the intensity in
the vertical direction within the divergence Qm of the incident beam of the instrument. That is, the smeared intensity
Ism,X(Q) for the parallel-cut sample is a line average given by
Ism;X Q 2Qm 1
IQ; q; 0 dq
Qm
Q
Rm
2Qm 1
Q
Rm
I Q2 q2
1=2
dq;
Qm
Figure 6
Figure 5
The ratio of the long- to short-axis intensities from the Z and Y curves,
1 with
plotted as a function of Q for the range 1 103 < Q < 1 102 A
a mean value 7.11 (14). Taking the exponent n = 3.33 (4), this indicates a
mean value for a/b = 1.80 (2).
938
Gu and Mildner
research papers
direction. Similarly, Z designates that the bedding plane is
placed in the vertical plane of the instrument, with the longer
axis of the elliptical contours in the horizontal direction.
In Appendix A we compare isocontours for the two
orientations of the perpendicular-cut sample with those for the
parallel-cut sample in order to determine the expected relative
intensities of the smeared USANS curves. The integral over
the slit height is greater when the bedding plane is in the
instrument horizontal direction, the Y measurement, than for
the parallel-cut sample. This is what is observed in Fig. 7 for
the smeared data, where the Y curve is no longer coincident
with the X curve for the parallel measurement as was the case
with the corresponding SANS measurements. Similarly, we
find that the Z measurement with the bedding plane in the
instrument vertical direction is reduced relative to the longeraxis SANS measurement. In addition, the curve for the
perpendicular-cut sample rotated in the neutron beam also lies
between the two, but is now closer to the Z curve.
We compare the four smeared USANS measurements in
Fig. 7. We performed straight-line fits to the loglog plot,
1,
assuming that, over the range 2 104 < Q < 2 103 A
m
the data are given by a power law Is(Q) = AQ . The results
are given in Table 1. Again, we find that, over this Q range, all
the gradients are similar, with the exponent m = 2.16 (4). The
large ratio (>5000) of the vertical to horizontal divergences
means that the integral in equation (8) extends over a large
range of q for all values of Q. Consequently, a power-law
scattering such as Qn becomes slit smeared as Qm, where
R2
0
Q
Rm
Qm
10
Figure 7
The slit-smeared USANS results plotted as a function of Qx, the
scattering vector in the horizontal high-resolution direction. The X curve
is for the parallel-cut sample, the Y and Z curves are for the
perpendicular-cut sample with the bedding plane in the horizontal and
vertical directions, respectively, and the R curve is for the rotation of the
perpendicular-cut sample.
J. Appl. Cryst. (2016). 49, 934943
Figure 8
The ratio of the smeared USANS intensities for the perpendicular-cut
sample with the bedding plane parallel and perpendicular to the
scattering plane, plotted as a function of Q for the range 4 105 <
1. The line is a hyperbolic fit to the data: IZ/IY = 2.31 (3)
Q < 2.5 103 A
3.33 (26) 105/Q.
Gu and Mildner
939
research papers
the others. This might suggest that, at correlation distances
much greater than 10 mm, the pores within the rock tend on
average to become less anisotropic. This could also be an
artifact of measurement close to the incident beam, and
indeed some researchers (King et al., 2015) discount data
1 on account of multiple scattering
below Q = 1 104 A
caused by thicker samples than used here. Whatever the
reason, this does not contradict the fact that the smaller scale
pores are anisotropic, as shown by the SANS results.
The separation of the Q dependence from the azimuthal
dependence can also be extended to the USANS data. Fig. 8
shows the ratio IZ/IY of the smeared intensity values for the
perpendicular-cut sample with the bedding plane parallel and
perpendicular to the scattering plane as a function of Q for the
1. Clearly the model has
range 4 105 < Q < 2.5 103 A
broken down, but this might be expected since the ratio a/b
must approach unity as the porosity becomes less anisotropic
at values of Q well beyond the SANS range, corresponding to
very large distances. The plot indicates that the ratio is a
hyperbola that may be fitted to IZ/IY = C Q0/Q, where C =
1. This suggests that
2.31 (3) and Q0 = 3.33 (26) 105 A
there is little evidence of loss of anisotropy in the pores within
this sample for a correlation distance at least of the order of
2/Q0 20 mm.
line fits to the loglog plot, assuming that, over the range 2
1, the data are given by a power law
104 < Q < 2 103 A
n
I(Q) = AQ . Again we find that, over this Q range, the
gradients are similar, with the exponent n = 3.15 (7). This is
close to the value of n = 3.33 (4) obtained for the SANS data
(see Table 1). Though it is not strictly valid we have also
desmeared the Y and Z data.
Fig. 9(a) shows all four of the desmeared USANS results on
the same plot as the SANS results. In the region around Q =
1 where the SANS and USANS data meet, there is a
0.001 A
smooth transition in scattering intensity for the symmetric
parallel-cut case X, and also for the rotation case R for the
perpendicular-cut sample which is quasi-symmetric. The
desmearing algorithm correctly produces a curve for the
azimuthally symmetric case that concatenates well with the
SANS results. On the other hand, we observe a shift in the
scattering intensity for both of the asymmetric perpendicularcut measurements. These abrupt changes in the combined data
can be better observed in Fig. 10(a), which shows the scattering intensity in the vicinity of the transition for both the Y
1.
and Z data at Q 0.001 A
We account for these changes in intensity at the crossover
point as follows. The desmearing algorithm requires that the
measured data be azimuthally symmetric, so that the scales of
Q in the horizontal scan axis and vertical integration axis are
commensurate, and the line averages are across the circular
contours. This procedure determines a smooth transition from
the SANS curve to the desmeared USANS result. The Y
SANS curve is from a sector average at the lowest intensity of
the two-dimensional data, whereas the Y USANS intensities
are derived from line averages in the integration direction
Figure 9
The combined results from the desmeared USANS (open square symbols) and the SANS (filled square symbols) measurements (a) before and (b) after
modification to the USANS data. The parallel-cut (brown X curve) and the rotational (green R curve) measurements show a smooth transition, whereas
for the perpendicular-cuts before modification the Y blue curve (with the bedding plane horizontal) has an upward jump at the transition and the Z red
curve (with the bedding plane vertical) has a downward jump. After modification, the transitions are no longer so abrupt.
940
Gu and Mildner
research papers
Figure 10
More detailed results from Fig. 9 in the vicinity of the overlap region of the SANS/USANS transition, showing (a) the upward jump for the Y curve and
the downward jump for the Z curve before modification, and (b) the less abrupt transition after modification. The SANS data have filled square symbols
and the USANS data have open square symbols. The line indicates the power law AQn for the X and Y curves, where n = 3.33.
2:05 1:762:16 :
11
AY b=rm AY b
This gives a value of 6.95 (24) for the intensity ratio, compared
with a value of 7.11 (14) for SANS shown in Fig. 5.
Note that modifying the intensity values by a given factor is
in effect the same as altering the sample thickness by the same
factor in the raw data correction. This begs a question
regarding the accuracy of the sample thicknesses. It affects
how well the data can be placed on an absolute basis, and
indicates the importance of the correct sample thickness.
6. Summary
We wish to modify USANS asymmetric data that follow a
power law such that they concatenate well with data collected
Gu and Mildner
941
research papers
using pinhole SANS. The ratio of the intensities for the Z and
Y SANS measurements depends on the aspect ratio a/b of the
elliptical scattering contours and on the power law exponent n:
AZ SANS an
:
12
b
AY SANS
The two measured smeared USANS intensities may each be
modified by the a/b ratio obtained from the SANS measurement to give
AZ USANS mod AZ USANS a=rm AZ USANS a=bm=2
13
14
and
so that
AZ USANS mod AZ USANS am
;
AY USANS mod AY USANS b
15
;
16
b
AY USANS b
so that
AZ USANS anm
:
b
AY USANS
17
7. Conclusions
We have considered the analysis of the simplest form of
azimuthal asymmetric scattering for measurements at scattering vectors below the normal SANS regime using a doubleperfect-crystal USANS instrument. We would like to extract
useful quantitative information from the measured data, for
which no mathematical model of the scattering system is
available. We have taken scattering measurements from thin
shale rock samples that are known to have narrow pores that
are slit like. These samples have been cut both parallel and
perpendicular to the plane of bedding. The normal to the
bedding plane is the unique axis that determines that the
scattering contours for the perpendicular-cut samples are
elliptical. The slit-smeared USANS results for the paralleland perpendicular-cut samples in the two different orientations have been given as a function of scattering vector. The
942
Gu and Mildner
APPENDIX A
We assume that, over the range of measurements, the
angular dependence of the scattering can be separated from
the scattering vector dependence. The isointensity contours
corresponding to the arrangement with the scattering plane
normal to the bedding plane are elliptical in shape, given by
equation (4), where is the azimuthal angle relative to the Z
direction. That is, tan = Q/q, where Q and q are the horizontal and vertical components of the scattering vector,
respectively. Fig. 11 shows two contours of the same intensity.
The circular contour for the X measurement (with the
parallel-cut sample) is tangential at the point (Q0 , 0) with the
elliptical contour for the Y measurement, with the long axis in
the vertical direction and normal to the high-resolution scan.
Since a > b, IY(Q, q) > IX(Q, q) for all q. The smeared intensity
Ism(Q) at a value of Q is given by equation (8). Hence, the
relative smeared intensities are given by Ism,Y(Q, q) >
Ism,X(Q, q).
Desmearing the USANS asymmetric data results in an
upward shift in intensity observed in the Y direction for the
perpendicular-cut sample relative to the X direction. (For the
SANS pinhole measurements, the intensities in the Y direction
are coincident with the X direction, the symmetric scattering
case.) Similarly, there is a downward shift observed at the
crossover point for the perpendicular-cut sample in the
direction Z normal to the bedding plane. This is in contrast
with both the X and R curves which concatenate smoothly.
J. Appl. Cryst. (2016). 49, 934943
research papers
Energy. This work utilized the NG3 SANS and BT5 USANS
instruments at the NCNR, which are supported in part by the
National Science Foundation under agreement No. DMR0944772. We thank Terry Engelder at Pennsylvania State
University for providing the shale samples. We also
acknowledge helpful discussions with John Barker.
References
Figure 11
A schematic diagram in reciprocal space indicating isointensity contours
of the same magnitude for the X (parallel-cut sample) measurement
(circular contour, shown in orange) and for the Y (perpendicular-cut
sample with the bedding plane horizontal) measurement (elliptical
contour, shown in blue). The contours are tangential at the point (Q0 , 0).
The path of the slit-smearing integral at constant Q is shown in green and
extends to Qm, corresponding to the maximum vertical divergence of
the incident beam. The intensity at (Q, q) is greater for the Y
measurement than for the X measurement for all Q, as is the integral
given by equation (8).
Acknowledgements
XG acknowledges support for the use of neutron scattering in
the analysis of rock porosity by the Department of Energy,
Basic Energy Sciences, grant No. DE-FG02-OSER15675 to
S. L. Brantley, and by the Laboratory Directed Research and
Development Program of Oak Ridge National Laboratory,
managed by UTBattelle LLC for the US Department of
Abbas, S., Desert, S., Brulet, A., Thevenot, V., Permingeat, P., Lavie,
P. & Jestin, J. (2015). J. Appl. Cryst. 48, 12421253.
Abramowitz, M. & Stegun, I. A. (1965). Handbook of Mathematical
Functions. Mineola, New York: Dover Publications.
Agamalian, M., Wignall, G. D. & Triolo, R. (1997). J. Appl. Cryst. 30,
345352.
Anovitz, L. M., Lynn, G. W., Cole, D. R., Rother, G., Allard, L. F.,
Hamilton, W. A., Porcar, L. & Kim, M.-H. (2009). Geochim.
Cosmochim. Acta, 73, 73037324.
Barker, J. G. (2015). Personal communication.
Barker, J. G., Glinka, C. J., Moyer, J. J., Kim, M. H., Drews, A. R. &
Agamalian, M. (2005). J. Appl. Cryst. 38, 10041011.
Bonse, U. & Hart, M. (1965). Appl. Phys. Lett. 7, 238240.
Brulet, A., Thevenot, V., Lairez, D., Lecommandoux, S., Agut, W.,
Armes, S. P., Du, J. & Desert, S. (2008). J. Appl. Cryst. 41, 161166.
Choi, S.-M., Barker, J. G., Glinka, C. J., Cheng, Y. T. & Gammel, P. L.
(2000). J. Appl. Cryst. 33, 793796.
Ciccariello, S., Schneider, J.-M., Schonfeld, B. & Kostorz, G. (2000).
Europhys. Lett. 50, 601607.
Dewhurst, C. D. (2014). J. Appl. Cryst. 47, 11801189.
Glinka, C. J., Barker, J. G., Hammouda, B., Krueger, S., Moyer, J. J. &
Orts, W. J. (1998). J. Appl. Cryst. 31, 430445.
Gu, X., Cole, D. R., Rother, G., Mildner, D. F. R. & Brantley, S. L.
(2015). Energy Fuels, 29, 12951308.
Gu, X., Mildner, D. F. R., Cole, D. R., Rother, G., Slingerland, R. &
Brantley, S. L. (2016). Energy Fuel. In the press.
King, H. E. Jr, Eberle, A. P. R., Walters, C. C., Kliewer, C. E., Ertas, D.
& Huynh, C. (2015). Energy Fuels, 29, 13751390.
Kline, S. R. (2006). J. Appl. Cryst. 39, 895900.
Lake, J. A. (1967). Acta Cryst. 23, 191194.
Matsuba, G., Ito, C., Zhao, Y., Inoue, R., Nishida, K. & Kanaya, T.
(2013). Polym. J. 45, 293299.
Milici, R. C. & Swezey, C. S. (2006). USGS Open-File Report 20061237.
Olsson, A., Hellsing, M. S. & Rennie, A. R. (2013). Meas. Sci.
Technol. 24, 105901.
Radlinski, A. P. (2006). Rev. Mineral. Geochem. 63, 363397.
Radlinski, A. P., Radlinska, E. Z., Agamalian, M., Wignall, G. D.,
Lindner, P. & Randl, O. G. (2000). J. Appl. Cryst. 33, 860862.
Reynolds, L. E. & Mildner, D. F. R. (1984). J. Appl. Cryst. 17, 411416.
Summerfield, G. C. & Mildner, D. F. R. (1983). J. Appl. Cryst. 16, 384
389.
Triolo, F., Triolo, A., Agamalian, M., Lin, J.-S., Heenan, R. K., Lucido,
G. & Triolo, R. (2000). J. Appl. Cryst. 33, 863866.
Gu and Mildner
943